Mutation is Biased Towards Fitness

@swamidass
In what ways are mutations biased towards beneficial ones? Can you give a reference please?

2 Likes

@Agauger so good to see you here again.

There are several examples of this. One very commonly known one is that transitions are simultaneously:

  1. More common than transversions. The cause of this is at least in part because of the biochemistry of mutations.
  2. More likely to be a conservative amino acid substitution. This cause of this at least in part is the structure of the of the genetic code.

Because fitness is intertwined with mutational bias in this case it is hard to untangle some key features of the data. An example of one attempt to do so is here: https://academic.oup.com/mbe/article/19/7/1022/1068565.

There are other examples too.

  1. Recombination allows for shuffling of a genome in a manner very likely not to cause harm, and increases (perhaps even dramatically) the chance of causing benefit.

  2. Copy number variation another example of a mutational mechanism that is not likely to cause harm (thought it sometimes can), but increases the chances of causing benefit.

  3. Mutational clusters localize point mutations together, which dramatically increases the chance of beneficial coordinated mutations in a protein. These were only recently discovered, and there are several classes. For some of them, we’ve established the bimolecular mechanisms, and expect to for the rest.

All these mutational mechanism are biased towards changes that are more likely to increase fitness. This list is not exhaustive. I could easily double it.

All this is widely agreed upon in the field. Though some specific examples are new, the general pattern has known for a long time, and is not unique to EES. Remember, I am not on board with EES, because this is all already part of evolutionary science.

3 Likes

Do these features of mutation strike you as unusual? Why should they be just so as to enhance beneficial mutations? Are we seeing these things because without them we would not be here? That strikes me as a weak anthropic principal sort of deal. Codes specially arranged, rearrangements in useful places etc. Or are we seeing what succeeded?7

4 Likes

They do seem to indicate life is arranged so as to evolve. Some if this can be explained as robustness to error, and some of it seems like necessary consequence of a rational world. Some of it, however, seems to be around almost exclusively for improving evolvability.

If there are two essentially identical populations, but one can evolve better because it has correctly biased it’s mutations, then the more evolvable poplution has a massive advantage. Even if the non evolvable poplution remains, the evolvable one will be able to extend into new environments and more. Evolvable appears to be a very beneficial trait, so it is not surprising to see it in life that arises by evolution.

This is only a weak anthropic signal if we are speaking from the context of common descent. Is that the context within which you are speaking?

Without common descent, there is much less reason to make life evolvable. If God specially created us, and our form was important to the image of God, he could have kept our form constant be removing our ability to evolve. Doing so would have also likely allow us to live free of cancer, which only exists because these evolabiltiy mechanisms are in our somatic cells too.

2 Likes

I would also add “rate of mutation” to the list. The fidelity of polymerases and other mechanisms seem to produce mutations at a rate that allows for evolution while preventing the species from being overwhelmed by detrimental mutations. This would also include “adaptive mutations” which is really an increase in the random mutation rate during times of stress.

3 Likes

I would agree with this. Good point.

Can you think of analogies in computer software coding or running computer systems where “random changes” to the underlying code result in better overall function?

Yes. Just about all of the software I write fits that category. That is how the most advanced machine learning algorithms work.

1 Like

The AlphaGo program still fascinates me. The game of Go was thought to be too complex for modern AI, but as it turns out the computers are now better at playing Go than the best humans. This was accomplished by essentially making random changes and keeping what worked best.

3 Likes

And would you say the software is designed that way? That is, does it function better now as the result of applied intelligence?

1 Like

Could this be due to harmful changes being selected-out more quickly? That is, the rate of beneficial mutations remains the same, but fewer harmful mutations are observed (because they do not survive).

So, it would seem that the analogy fits? Would this be an “expected feature” of a software code which was written on a faulty computer, by a toddler hacking the keyboard with a stick, which also accidentally and randomly substituted terms and/or functions in the code?
If you say, “unlikely, but not completely impossible,” I’ll grant you brownie points for candor. Or, you could just remain silent on the question and ignore me…

Not really. However, it should be noted that when recombination is blocked, more severe trisomies and monosomies are caused by nondisjunction, as recombination is part of the mechanism for lining up homologous pairs of chromosomes during meiosis.

2 Likes